Robust Transmission for Energy-Efficient Sub-Connected Active RIS-Assisted Wireless Networks: DRL Versus Traditional Optimization

Abstract

This paper investigates the performance of a sub-connected active reconfigurable intelligent surface (RIS)-assisted communication system under imperfect channel state information (CSI). To ensure reliable transmission, we formulate an optimization problem aimed at maximizing the energy efficiency (EE) of the system. This optimization problem involves the joint optimization of the transmit precoder at the base station (BS) and the beamforming matrix at the RIS while considering a norm-bounded CSI error model. Given the non-convex nature of this problem, we employ deep reinforcement learning (DRL)-based methods, including deep deterministic policy gradient (DDPG), proximal policy optimization (PPO), and modified PPO, to find the optimal transmit precoder and beamforming matrix ensuring an energy-efficient operation. Additionally, we introduce an analytical framework to address this problem using traditional analytical optimization (TAO) techniques. Through extensive simulations, we showcase the convergence, robustness, and effectiveness of the proposed algorithms when compared to TAO-based solutions. Furthermore, we also highlight the impact of various system parameters, such as the total number of elements, the required number of amplifiers, and the maximum available transmit power at the BS, on the performance of the examined communication system.